Comparative genomic hybridization was applied to 5 breast cancer cell lHes and 33 primary tumors to discover and map regions of the genome with increased DNAsequence copy-number. Two-thirds of primary tumors and almost all cell lines showed increased DNA-sequence copynumber affecting a total of 26 chromosomal subregions. Most of these loci were distinct from those of currently known amplified genes in breast cancer, with sequences originating from 17q22-q24 and 20q13 showing the highest frequency of amplification. The results indicate that these chromosomal regions may contain previously unknown genes whose increased expression contributes to breast cancer progression.Chromosomal regions with increased copy-number often spanned tens of Mb, suggesting involvement of more than one gene in each region.Increased expression of specific genes plays an important role in the pathogenesis of solid tumors (1-3). Gene amplification, characterized by distinct cytogenetic structures, such as homogeneously stained regions, double-minute chromosomes (1-7), is commonly found in tumor cells and is considered an important mechanism by which tumor cells gain increased levels of expression of critical genes. Increased copy-numbers also occur as a result of extensive chromosomal rearrangements, such as duplications, isochromosomes, extra marker chromosomes, and acentric chromosomal fragments that may affect the gene dosage of numerous genes simultaneously. In breast cancer, cytogenetic evidence of increased DNA-sequence copy-number is common (4-7). For example, homogeneously stained regions have been found in 60% of primary breast carcinomas (7). Although genetic analysis has found amplification of oncogenes, such as ERBB2 (17q12), MYC (8q24), PRADII CYCLIN D (11q13), FLG (8p12), BEK (10q24), and IGFR-1/FES (15q24-q25) (8-12), in most cases these do not explain the presence of large homogeneously stained regions (13). Thus, amplification of currently unknown genes may often occur in breast cancer.We have recently developed a method, comparative genomic hybridization (CGH), for surveying entire genomes for DNA-sequence copy-number variation (14, 15). In CGH, the relative intensities of tumor DNA (detected using green fluorescence) and normal reference DNA (detected with red fluorescence) after hybridization to normal metaphase chromosomes is used to reveal and map regions of increased DNA-sequence copy number (14-16). These loci are visualized as chromosomal region(s) with predominantly green fluorescence ( Fig. 1) and quantified by digital image analysis as an increased green-to-red fluorescence intensity ratio (Fig. 2). As no specific probes or previous knowledge of aberrations is required, CGH is especially suitable for identification and mapping of previously unknown DNA copy-number changes that may highlight locations of important genes. In the present study, we have used CGH to identify and map increases in DNA-sequence copy number in 15 breast cancer cell lines and 33 uncultured primary breast tumors. MATERIALS AND ME...

Loss of heterozygosity (LOH) was detected in morphologically normal lobules adjacent to breast cancers. The most frequent aberration was at chromosome 3p22-25; of ten cases with this LOH in the carcinoma, six displayed the same LOH in adjacent normal lobules. This suggests that in a subset of sporadic breast cancers, a tumor suppresser gene at 3p22-25 may be important in initiation or early progression of tumorigenesis. Among sixteen breast cancers with LOH at 17p13.1 and five breast cancers with LOH at 11p15.5, one case each displayed the same LOH in adjacent normal lobules. Thus the molecular heterogeneity that characterizes invasive breast cancers may occur at the earliest detectable stages of progression.The mature breast contains lobules, clusters of closed glandular spaces that produce milk during lactation. These lobules are connected to the nipple-areolar complex by a system of branching ducts that are surrounded by varying amounts of fat and connective tissue. Breast cancer is thought to develop within a terminal ductal-lobular unit (TDLU), which includes the lobule and its most proximal ducts (1).Breast cancer evolves by clonal selection of cells that acquire multiple molecular changes. One model suggests that breast cancer, like colon cancer (2), develops through a defined progression of morphologically distinguishable stages beginning with benign hyperplasia, which progresses to atypical hyperplasia, then to in situ carcinoma, and finally to invasive cancer (1). This sequential progression may not be the only way that breast cancers develop, however. Many small invasive cancers do not have atypical components, which suggests that they may have developed directly from morphologically normal epithelium. If this were true, one might expect to find evidence of a "field effect" in which at least some of the genetic aberrations found in invasive cancers are also present in the morphologically normal epithelium.To test this hypothesis, we carefully microdissected hematoxylin-eosin-stained sections of breast cancers so as to isolate morphologically discrete regions (Fig. 1A). DNA was prepared from malignant areas of the section and from adjacent normal TDLUs. As a control for each case, DNA was also prepared from normal breast skin (usually from a separate section) that had been similarly microdissected.We studied LOH at chromosome 3p24, 11p15.5, 13q13, and 17p13.1 because these loci show LOH in a high percentage (ϳ30 to 60%) of invasive ductal breast cancers (3, 4). For the carcinomatous regions, the frequency of LOH at 3p24 (48%) and 11p15.5 (29%) was similar to that previously reported (4). The frequency of LOH in the invasive components was higher than the literature values for 13q13 (64% here versus ϳ40%) and for 17p13.1 (80% here versus ϳ60%). These discrepancies may be due to random variation because our sample size was small.In 8 of 30 cases we detected LOH in the adjacent morphologically normal TDLUs (Table 1). In all eight cases, the same allele was missing in the adjacent carcinoma (Fig. 2, A an...

We illustrate the use of fluorescence in situ hybridization (FISH) for analysis of ERBB2 oncogene copy number, the level of amplification (here defined as the ratio of ERBB2 copy number to copy number of chromosome 17 centromeres), and the distribution of amplified genes in breast cancer cell lines and uncultured primary breast carcinomas. The relative ERBB2 copy number determined by FISH in 10 breast cancer cell lines correlated strongly with Southern blot results (r = 0.98) when probes for an identical reference locus were used in the two methods. Metaphase analysis of cell lines showed that amplified ERBB2 copies always occurred in intrachromosomal clusters but that the number and chromosomal location of these clusters varied among the cell lines. In interphase nuclei of primary tumors showing ERBB2 amplification (10/44), ERBB2 copies were seen as one to four clusters, also suggesting intrachromosomal localization. Regardless of the average level of amplification, all these tumors contained highly amplified cell subpopulations with at least 25, and sometimes more than 100, ERBB2 copies per cell. Tumors that did not show amplification by FISH (34/44) had an average of one to five ERBB2 copies scattered randomly in the nuclei and completely lacked cells with high copy levels. FISH results on primary tumors were concordant with slot blot results on amplification and with immunohistochemical detection ofoverexpression. Quantitative analysis of ERBB2 amplification by FISH may improve prognostic assessments based on the pattern of amplification and detection of heavily amplified tumor cell subpopulations.